Assessment of Polarimetric Parameters in L-Band SAR for Land Cover Classification
Entezari, Iman1; Motagh, Mahdi2; Babak, Mansouri3
1University of Alberta, CANADA; 2GFZ German Research Centre for Geosciences, GERMANY; 3International Institute of Earthquake Engineering and Seismology, IRAN, ISLAMIC REPUBLIC OF
This research presents an investigation on the use of the fully polarimetric SAR data for land cover classification and identification. The main goal of this research is to assess the potential of various polarimetric parameters in land cover identification and the information that they contain in terms of sensitivity to the different land types.
SAR complex analysis and target decomposition techniques are employed to derive the polarimetric descriptors including backscattering cross sections, ratios of scattering matrix elements, polarimetric coherencies, Pauli coefficients, Krogager coefficients, Freeman coefficients, Yamaguchi coefficients, H/A/Alpha parameters, and eigenvalues of the coherency matrix. These parameters are then employed in classic supervised Maximum Likelihood, Minimum Distance, and Parallelepiped classifiers.
In order to assess the results, fully polarimetricL-band ALOS-PALSAR data from Tehran is used. Results show that although the performance of the classification tests depends on the type of classifier and selection of the polarimetric parameters, the polarimetric parameters extracted from target decomposition techniques in Maximum Likelihood Classifier generally lead to a high accuracy in land cover classification. Furthermore, the polarimetric parameters derived from SAR complex analysis are more effective in circular basis than in linear basis in terms of classification accuracy.